A very interesting question popped up on Flickr: Is there any technical advantage of specifying AdobeRGB or sRGB in a Camera RAW file?
Almost all digital cameras obtains color by placing a color filter in front of identically constructed photodiodes. The RAW file just stores digitization of those monochromatic values. Color profile information on a RAW file is stored in the metadata, not applied to the file itself. So there is no technical difference between the two profiles besides hinting to your image processing applications your preferred color space.
There is a slight theoretical exception here: the RAW format isn’t a standard so there is no reason why a camera manufacturer couldn’t record different raw data based on the intended color space. This might be advantageous in a camera like the Nikon D200, which does color processing in analog space before digitization. The intended color space could theoretically provide hints to the camera so as to minimize interpolation to the resulting color space.
Of course, this doesn’t happen. If Nikon did such a thing, Adobe would probably have a conniption and call it “encryption.” Sometimes I wonder if Nikon’s White Balance code page in the D200 was intended to give Nikon engineers the flexibility to take advantage of this camera trait—I know of no other camera that uses white balance information to modify digitization—in a future firmware release…and corresponding Capture NX update.
Now we’ll never know.
[Adobe Camera RAW and color profiling after the jump]
Adobe Camera RAW and color profiling
It seems all the time I’m unnecessarily nasty to Adobe. Guilty!
In spite of this, I like to point out an excellent article I found while researching this answer.
The article explains how to use Adobe Camera RAW (in Photoshop CS and later) to make a custom calibrated profile for your camera. This is an excellent feature I was not aware of and may be essential for certain studio photographers. It also explains why ACR uses separate profiling for tungesten vs. daylight lighting—very interesting and informative and it increases my respect for how Adobe engineers are both exacting and pioneering when it comes to color reproduction.
(In my anti-Adobe bias, I will take exception to the article’s generalization that vendor-supplied RAW processors are poor. Perhaps the author shoots Canon and likes overly-contrasty images. Nikon Capture and Capture NX is superior to ACR in stability and accuracy, though not speed. Olympus processing has incorporated metadata lensfix data in Camedia since its inception. It seems that such a comment should be considered on a vendor-by-vendor basis.)
3 thoughts on “RAW color space”
In the thread starfish235 points out that loss can occur in DNG files when flipping between colorspaces. Interesting.
Also he notes that Nikon Capture 4 has poor UI for flipping color space.
"I know of no other camera that uses white balance information to modify digitization"
The D300 definitely does, but maps back to D65 before saving so it's harder to see the quantization effects than it is with the D200. Wintrhawk took a series of D300 photos of the same scene with different WB settings and we were able to pull out the digitization / quantization effects. The mapping back to D65 seems to be the new Nikon standard.
Just a thought from a starfish